UniMMVSR: A Unified Multi-Modal Framework for Cascaded Video Super-Resolution

Shian Du^{1 †}, Menghan Xia^{2 ✉}, Chang Liu¹, Quande Liu³, Xintao Wang³, Pengfei Wan³, Xiangyang Ji^{1 ✉
1}Tsinghua University ²Huazhong University of Science and Technology ³Kling Team, Kuaishou Technology
†: Intern at Kuaishou Technology, ✉: Corresponding Authors

📋 News

• [2025.10.10] Release Arxiv paper.

📖 Introduction

We propose UniMMVSR, the first unified generative video super-resolution framework to incorporate hybrid-modal conditions, including text, images, and videos, which supports 4K controllable video generation for the first time.

⚙️ Code (Coming soon)

Citation

 @article{du2025unimmvsr,
  title={UniMMVSR: A Unified Multi-Modal Framework for Cascaded Video Super-Resolution},
  author={Du, Shian and Xia, Menghan and Liu, Chang and Liu, Quande and Wang, Xintao and Wan, Pengfei and Ji, Xiangyang},
  journal={arXiv preprint arXiv:2510.08143},
  year={2025}
}